Digital Video Broadcasting By Satellite

Digital Video Broadcasting By Satellite Matthew C. Valenti Lane Department of Computer Science and Electrical Engineering West Virginia University U.S.A. Apr. 2, 2012 ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 1 / Univer 41

Acknowledgements I would like to thank: Xingyu Xiang. National Science Foundation. Army Research Lab. Hughes Network Systems. DirecTV. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 2 / Univer 41

Outline 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 3 / Univer 41

Outline Satellite Television Standards 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 4 / Univer 41

Satellite Television Standards Providers Digital Satellite Television in the United States DirecTV Spinoff of Hughes Network Systems. Began operations in 1994. 12 geosynchronous satellites. 20 million U.S. subscribers at end of 2011. 23,000 employees in U.S. and Latin America. $33.4 billion market cap. Dish Network. Spinoff of EchoStar. Began operations in 1996. 14 million U.S. subscribers in at end of 2011. 22,000 employees. $14.7 billion market cap. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 5 / Univer 41

Satellite Television Standards Providers The DVB Family DVB is a family of open standards for digital video broadcasting. Maintained by 270-member industry consortium. Published by ETSI. Modes of transmission Satellite: DVB-S, DVB-S2, and DVB-SH Cable: DVB-C, DVB-C2 Terrestrial: DVB-T, DVB-T2, DVB-H ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 6 / Univer 41

DVB-S Satellite Television Standards Providers.6'."AB3C"97)(@?F"G).6".6)?">$7H";'%-".7'(?%)??)$("=';.$7?"?6$+E,"9-".'H-(")(,-?)@(F"I'.6-7".6'(";$(?),-7)(@"$(E8".6.6-"&7-<)$+?"/0123"'(,"/012/3J*"?&-?&-;)=);'.)$("7-;$@()?-?".6-"=$EE$>)(@"-=!"#$%&'()$%& *('+,)'%#-&.#/')012%&#'2&3"#$%&4"#-#4 6"%&5#40&0"#0&0"%&0-#'$3('2%-&)$&3(7%-&,). 8-(13&2%,#9&%55%40$& 56)?">$7H"E-,".$".6-",-=)(-,";$(?.-EE'.)$ ;$(,).)$(?F"56-";$(?.-EE'.)$(?".6'.">-7- ; : ;!<= ; Modulation: QPSK with α = 0.35 rolloff. Channel coding: Concatenated Reed Solomon and convolutional.?@a!<= : ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West "#$%&'!()!*+,-./!01234'55 2012Virginia 7 / Univer 41

Satellite Television Standards Providers DVB-S2 DVB-S2 was introduced in 2003 with the following goals: Improve spectral efficiency by 30% through better modulation and coding. Modulation: QPSK, 8PSK, 16/32 APSK. Channel coding: LDPC with outer BCH code. Offer a more diverse range of services. HDTV broadcast television. Backhaul applications, e.g., electronic news gathering. Internet downlink access. Large-scale data content distribution, e.g., electronic newspapers. Ratified 2005. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 8 / Univer 41

Satellite Television Standards Adaptive Internet Downlink Providers!!"#$%&'()$("*+),-".$"/01234! >;.9$3?))5 @&*A,*5$/,(,$8,(9 /01234 #&5'6,(&* >;.9$3?))5 @&*A,*5$/,(,$8,(9 738 7-()*-)( %) ('*- "9,--)6!"#$%&'()*+ #,-,.)* %:$3;.-,6 <',6;(= %) ('*- "9,--)6 $ "#$%&'!(()!*+!*,-!./01'2! ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012Virginia 9 / Univer 41

Outline DVB-S2 Modulation 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 10 / Univer 41

DVB-S2 Modulation Why Not Use QAM? A constellation with M = 2 m conveys m bits per channel use. Higher spectral-efficiencies require larger signal constellations. Nonlinear satellite channels are not well suited to square QAM. Nonlinear! TWTA! ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 11 / Univer 41

Non-linear magnitude DVB-S2 Modulation and phase characteristics of a saturated transponder The DVB-S2 Signal Constellations The fact that the transponder is power limited Group delay effects This work led to the defined constellations to be optimised for the above conditions. The constellations that were chosen are shown below: Q Q I I QPSK Q 8PSK Q I I 16APSK 32APSK Figure 1: DVB-S2 Constellations ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 12 / Univer 41

DVB-S2 Modulation Raised-Cosine Rolloff Filtering DVB-S2 uses a tighter root RC-rolloff filter. B = R s (1 + α) 1.25 Assuming a 6 MHz 1 transponder channel... DVB-S Example: 0.75 α = 0.35. QPSK: R b = 2R s 0.5 2(6)/(1.35) = 8.9 Mbps = 0.20 DVB-S2 Example: = 0.25 0.25 = 0.35 α = 0.20. 32-APSK: R b = 5R s 0 5(6)/(1.2) = 25 Mbps 0.75 0.5 0.25 0 0.25 0.5 0.75 H(f) 2f/R s ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 13 / Univer 41

DVB-S2 Modulation Uncoded BER in AWGN 10 0 10 1 32APSK 16APSK 8PSK QPSK 10 2 BER 10 3 10 4 10 5 10 6 0 5 10 15 20 25 Es/No in db ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 14 / Univer 41

Outline LDPC Coding 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 15 / Univer 41

LDPC Coding Maximum Information Rate Performance can be improved by using error control coding. Gains are limited by the modulation-constrained capacity. Capacity (bits per channel use) 6 5 4 3 2 1 32APSK 16APSK 8PSK QPSK LDPC codes are capable 0 of approaching capacity. 10 5 0 5 10 15 20 25 Es/No in db Symmetric information rate (assumes uniform input). ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 16 / Univer 41

LDPC Coding Available Code Rates The encoder maps length-k messages to length-n codewords. The code rate is R = k/n. Useful bit rate is R u = R log 2 (M). Two codeword lengths: 16, 200. 64, 800. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 17 / Univer 41

LDPC Coding Coding by Example: Single Parity-Check Codes Consider the following rate R = 5/6 single parity-check code: c = [1 0 1 0 1 1 ] } {{ } } {{ } u parity bit One error in any position may be detected: c = [ 1 0 X 0 1 1 ] Problem with using an SPC is that it can only detect a single error. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 18 / Univer 41

LDPC Coding Product Codes Place data into a k by k rectangular array. Encode each row with a SPC. Encode each column with a SPC. Result is a rate R = k 2 /(k + 1) 2 code. Example k = 2. c 1 = u 1 c 2 = u 2 c 3 = c 1 c 2 c 4 = u 3 c 5 = u 4 c 6 = c 4 c 5 = c 7 = c 1 c 4 c 8 = c 2 c 5 c 9 = c 3 c 6 1 0 1 1 1 0 0 1 1 A single error can be corrected by detecting its row and column location 1 0 1 0 1 0 0 1 1 1 0 1 1 1 0 0 1 1 ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 19 / Univer 41

LDPC Coding Linear Codes c 1 = u 1 c 2 = u 2 c 3 = c 1 c 2 c 4 = u 3 c 5 = u 4 c 6 = c 4 c 5 c 7 = c 1 c 4 c 8 = c 2 c 5 c 9 = c 3 c 6 The example product code is characterized by the set of five linearly-independent equations: c 3 = c 1 c 2 c 1 c 2 c 3 = 0 c 6 = c 4 c 5 c 4 c 5 c 6 = 0 c 7 = c 1 c 4 c 1 c 4 c 7 = 0 c 8 = c 2 c 5 c 2 c 4 c 8 = 0 c 9 = c 3 c 6 c 3 c 6 c 9 = 0 In general, it takes (n k) linearly-independent equations to specify a linear code. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 20 / Univer 41

LDPC Coding What is a Parity-Check Matrix? The system of equations may be expressed in matrix form as: where H is a parity-check matrix. Example: c 1 c 2 c 3 = 0 c 4 c 5 c 6 = 0 c 1 c 4 c 7 = 0 c 2 c 4 c 8 = 0 c 3 c 6 c 9 = 0 System of equations ch T = 0 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 H = 1 0 0 1 0 0 1 0 0 0 1 0 1 0 0 0 1 0 0 0 1 0 0 1 0 0 1 Parity-check matrix ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 21 / Univer 41

LDPC Codes LDPC Coding An LDPC code is a code with a large, sparse H matrix. A code from MacKay and Neal (1996): 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 H = 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 The code called a (3, 4) regular code because: Each column has exactly 3 ones. Each row has exactly 4 ones. The DVB-S2 LDPC codes are irregular. More about this later. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 22 / Univer 41

DVB-S2 vs. Shannon LDPC Coding ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 23 / Univer 41

Outline Tricks for Improving Performance 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 24 / Univer 41

Tricks for Improving Performance BICM-ID Iterative Demodulation and Decoding Conventional receivers first demodulation, then decode. LDPC Decoder Hard decision Performance is improved by iterating between the demodulator and decoder. y APSK demodulator Π 1-1 VND _ Π 3-1 CND BICM-ID: bit-interleaved modulation with iterative decoding. L a (z) Π 1 Feedback LLRs Π 3 ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 25 / Univer 41

BICM vs. BICM-ID Tricks for Improving Performance BICM-ID 10 0 10-1 10-2 BER 10-3 10-4 10-5 Rate 4 BICM (4by5 LDPC) Rate 4 BICM-ID (4by5 LDPC) Rate 3.75 BICM (3by4 LDPC) Rate 3.75 BICM-ID (3by4 LDPC) Rate 3 BICM (3by5 LDPC) Rate 3 BICM-ID (3by5 LDPC) 10-6 9.8 10.1 10.4 10.7 11 11.3 11.6 11.9 12.2 12.5 12.8 13.1 13.4 13.7 E s /N 0 in db Curves show performance of 32APSK in AWGN. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 26 / Univer 41

Tricks for Improving Performance Constellation Shaping Constellation Shaping The symmetric information rate curves assume equiprobable signaling. It is possible to reduce the energy required to achieve a certain information rate by transmitting higher-energy signals less frequently than lower-energy signals. 2 1.5 1 0.5 0-0.5-1 -1.5-2 -2-1 0 1 2 Figure: Uniform 32APSK o vs. shaped 32APSK o. Both constellations have the same energy. mutual information 5 4 3 2 1 uniform shaping g=1 11.211.411.611.8 12 0-10 0 10 20 30 Es/No (db) Figure: The capacity of shaped 32APSK is about 0.3 db better than uniform 32APSK. 4 3.9 3.8 ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 27 / Univer 41

Tricks for Improving Performance Constellation Shaping 01 Sub-constellations The 32APSK is partitioned into two equal-sized sub-constellations. A shaping bit selects the sub-constellation, while the remaining bits select a symbol from the chosen sub-constellation. The lower-energy sub-constellation is selected more frequently. 01100 11101 10100 01101 10101 00100 00101 01111 11100 10000 00000 11000 01000 10001 11001 00001 01001 00111 11111 00110 00010 10111 01110 10110 10010 00011 10011 01011 11011 11110 11010 01010 Figure: 32APSK w/ 2 sub-constellations Figure: 32APSK symbol-labeling map ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 28 / Univer 41

Tricks for Improving Performance Constellation Shaping Shaping Encoder The shaping encoder should produce more zeros than ones. Example: (n s, k s ) = (5, 3) 3 input data bits 5 output codeword bits 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0 0 0 1 0 0 1 1 0 0 1 0 0 1 0 0 0 1 0 0 0 1 0 1 1 0 0 0 0 1 1 0 0 0 0 1 1 1 1 1 1 0 1 0 0 p 0 = 31/40: fraction of zeros. p 1 = 9/40: fraction of ones. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 29 / Univer 41

Tricks for Improving Performance Receiver Implementation Constellation Shaping Demodulator + Shaping Decoder LDPC Decoder Hard decision i y APSK demodulator S/P 2-1 shaping decoder P/S 1 _ 3 1-1 VND -1 CND L a (z) P/S 2 S/P 1 3 Feed dback LLRs Additional complexity relative to BICM-ID due to shaping decoder. MAP shaping decoder compares against all 2 ks shaping codewords. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 30 / Univer 41

Tricks for Improving Performance BER of Shaping in AWGN Constellation Shaping 10 0 10-1 BICM Uniform BICM-ID Uniform Shaping (4,2) Shaping (6,3) Shaping (12,6) 10-2 BER 10-3 10-4 10-5 10-6 4.4 4.5 4.6 4.7 4.8 4.9 5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 6 E b /N 0 in db BER of 32-APSK in AWGN at rate R=3 bits/symbol. Code rates: R c = 3/5 for uniform and R c = 2/3 for shaped. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 31 / Univer 41

Tricks for Improving Performance Degree Optimization Irregular vs. Regular Codes An irregular LDPC code has columns with different weights. The Hamming weight of a column is the number of 1 s. An irregular code can outperform a regular code. The main design consideration is the degree distribution, which quantifies how many columns there are of a particular weight. The optimal degree distribution can be found through linear programming. The optimization takes into account the APSK modulation and shaping (if used). ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 32 / Univer 41

Tanner Graphs Tricks for Improving Performance Degree Optimization The parity-check matrix may be represented by a Tanner graph. Bipartite graph: Check nodes: Represent the n k parity-check equations. Variable nodes: Represent the n code bits. If H i,j = 1, then i th check node is connected to j th variable node. Example: For the parity-check matrix: 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 H = 1 0 0 1 0 0 1 0 0 0 1 0 1 0 0 0 1 0 0 0 1 0 0 1 0 0 1 The Tanner Graph is: ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 33 / Univer 41

Tricks for Improving Performance Degree Optimization Degree Distribution Edge-perspective degree distributions: ρ i is the fraction of edges touching degree i check nodes. λ i is the fraction of edges touching degree i variable nodes. For example, consider the Tanner graph: 15 edges. All are connected to degree-3 check nodes, so ρ 3 = 15/15 = 1. Four are connected to degree-1 variable nodes, so λ 1 = 4/15. Eight are connected to degree-2 variable nodes, so λ 2 = 8/15. Three are connected to the degree-3 variable node, so λ 3 = 3/15. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 34 / Univer 41

Tricks for Improving Performance Degree Optimization Optimal Degree Distributions First, let s optimize for uniform 32-APSK. The DVB-S2 standard rate R c = 3/5 LDPC code has degree distributions: ρ 11 = 1 λ 2 = 0.182 λ 3 = 0.273 λ 12 = 0.545 The optimized degree distributions are: ρ 11 = 1 λ 2 = 0.182 λ 4 = 0.473 λ 19 = 0.345 A similar optimization can be performed for shaped 32-APSK. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 35 / Univer 41

Tricks for Improving Performance Degree Optimization BER with Optimized Degree Distributions 10-1 10-2 BICM-ID Uniform BICM-ID Uniform with optimized 3/5 LDPC code DVB-S2 2/3 LDPC and (4,2) shaping code optimized 9/14 LDPC and (3,2) shaping code 10-3 BER 10-4 10-5 10-6 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6 E b /N 0 in db BER of 32-APSK in AWGN at rate R=3 bits/symbol. Comparison of standard vs. optimized LDPC codes. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 36 / Univer 41

Tricks for Improving Performance Summary of Performance Gains Cumulative Gains 10-1 10-2 10-3 BICM Uniform BICM-ID Uniform BICM-ID Uniform with optimized 3/5 LDPC code DVB-S2 2/3 LDPC and (4,2) shaping code optimized 2/3 LDPC and (4, 2) shaping code optimized 9/14 LDPC and (3,2) shaping code BER 10-4 10-5 10-6 4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 E b /N 0 in db 0.33 db gain from using BICM-ID. 0.46 db gain from using shaping. 0.34 db gain from optimizing LDPC code. Cumulative gain of 1.13 db. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 37 / Univer 41

Outline Conclusion 1 Satellite Television Standards 2 DVB-S2 Modulation 3 LDPC Coding 4 Tricks for Improving Performance 5 Conclusion ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 38 / Univer 41

Conclusion Conclusion DVB-S2 is a highly efficient system, thanks to APSK modulation. Tight RC-rolloff filtering. Capacity-approaching irregular LDPC codes. The performance of DVB-S2 can be improved by BICM-ID. Constellation shaping. Optimization of LDPC degree-distribution. The cumulative gain is 1 db with all of these. Future work: Application to 64APSK and other modulations. Joint design of shaped modulation and code. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 39 / Univer 41

References Conclusion 1 and X. Xiang, Constellation shaping for bit-interleaved coded APSK, in Proc. IEEE Int. Conf. on Commun. (ICC), (Kyoto, Japan), June 2011. 2 C. Nannapaneni,, and X. Xiang, Constellation shaping for communication channels with quantized outputs, in Proc. Conf. on Info. Sci. and Sys. (CISS), (Baltimore, MD), Mar. 2011. 3 X. Xiang and, Improving DVB-S2 performance through constellation shaping and iterative demapping, in Proc. IEEE Military Commun. Conf. (MILCOM), (Baltimore, MD), Nov. 2011. 4 X. Xiang and, LDPC-coded APSK with constellation shaping and optimized degree distributions, submitted to IEEE Globecom, (Anaheim, CA), Dec. 2012. In review. 5 and X. Xiang, Constellation shaping for bit-interleaved LDPC coded APSK, submitted to IEEE Trans. Commun., revision under review. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 40 / Univer 41

Conclusion Thank You. ( Lane Department LDPCof Codes Computer Science and Electrical Engineering Apr. 2, West 2012 Virginia 41 / Univer 41